When I tried to turn on gpg encryption, encryption to s3 seems to work fine, but fetching/decryption fails. What ends up happening is there is are two files, an intermittent, encrypted file and the final file name which is 0 byte length. If I manually run gpg on the cmdline with the right key it will do the right thing. Why is s3cmd bailing on its call to gpg? How can I debug this?
What s3cmd and Python versions do you use? There was a bug with 0.9.9-pre5 and earlier running under Python 2.6 that may have caused this behaviour. Please upgrade to 0.9.9-rc3 and let me know if you still experience problems.
This happened to me on Fedora Core 8 using s3cmd from the s3cmd repos. It also now happens to me using the s3cmd that comes with debian lenny:
$ s3cmd --version
s3cmd version 0.9.8.3
The problem is NOT that gpg is not working -- it is. When I issue s3cmd get with a previously encrypted file, the file is downloaded from S3 with the same file name. Then gpg goes to work decrypting to a file with a random hash appended to the file. This file is properly decrypted. What fails is that the properly decrypted file does not then overwrite the originally downloaded, still encrypted file.
Is it still happening in 0.9.9? If not - are there any errors reported?
s3cmd 0.9.9 should work just fine in FC8 - simply download the tarball, unpack and run. No installation needed.